Somewhere beneath the Americas there is a sizeable volume of what formerly constituted the East Pacific ocean lithosphere. It represents half the productivity of the East Pacific Rise over more than 100 Ma. Although there is still considerable uncertainty about where such subducted rugs end up, seismic tomography does suggest that a fair proportion may reach the core-mantle boundary. That region of the mantle also seems to be the source of at least some mantle plumes. So it would not be very surprising if lavas formed from some plumes carried a signature from much older lithosphere. Finding such signs is not so easy, but if one pops out of lava geochemistry it would indicate that mantle convection has not stirred up and chemically blended the mess of subducted material in the lower mantle; a “memory” of bygone tectonics. At least 3 billion years of plate tectonics has contributed to the geochemistry of the mantle, so finding such a memory has been just a matter of patience, developing a means of teasing it out and luck.
One such signature has emerged from the plume-related islands volcanic islands of the Azores, in the form of an anomalously low 187Os/188Os isotopic ratio (Schaefer, B.F. et al. 2002. Evidence for recycled Archaean oceanic mantle lithosphere in the Azores plume. Nature, v. 420, p. 304-307). The study shows that the parent isotope (187Re) was depleted in the Azores source mantle up to 2500 Ma ago, perhaps before. Rhenium depletion is likely to occur in mantle rocks during partial melting, because it is incompatible, while osmium is compatible with mantle mineral assemblages that constitute the residue of melting. So the most likely explanation for unusually low 187Os is that oceanic mantle lithosphere, depleted by late-Archaean melting events, has sat around somewhere without being blended with more primitive mantle. Lead isotopes in modern ocean-floor basalts suggest that recycling on timescales around 2 billion years has occurred, and the Os data from the Azores confirm that. However, this is the first swallow in what may (or may not) become an osmium-isotope summer for geochemists eager to map the mantle’s evolution. And there is one big question: from what depth did the Azores plume rise? There is absolutely no evidence for it having risen from the core-mantle boundary (or anywhere else for that matter). So all the data really show is that Archaean materials have been incompletely mixed with their mantle surroundings. They could be products of Archaean subduction, but it requires special pleading to remove the possibility of Archaean lithosphere that resided just beneath the African or American continents before the Atlantic Ocean began to form.
Beowulf and mapping the mantle
Seismic tomography is a child of high-speed computing, of which we could barely dream only 10 years ago, as well as the world-wide network of seismic stations set up to detect nuclear tests. The grist to its mill is seismographic data supplied near instantaneously by modern broadband data telemetry. Mathematically it is not an easy subject, so an insight into how it is done is very welcome (Komatitsch, D. et al. 2002. The spectral-element method, Beowulf computing, and global seismology. Science, v. 298, p. 1737-1742). “Beowulf” refers to the use of clusters of ordinary PCs to perform the calculations, rather than single, main-frame supercomputing. The review outlines the theoretical approach of the spectral-element method (still beyond me!), but is most interesting in assessing the potential of future machines able to operate 100 times faster (petaflop machines) than even the most powerful today. It begins to look like geophysicists will unveil far more complexity in the mantle than geochemists have been able to sift from their analyses of exposed rocks at the surface.